Pith. sign in

REVIEW 4 major objections 4 minor 38 references

Next-Generation Phishing: How LLM Agents Empower Cyber Attackers

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that current phishing defenses—commercial filters, classical machine-learning classifiers, and LLM-based detectors—lose substantial accuracy against phishing emails rewritten by LLMs, and that retraining on LLM-augmented…

desk verdict The paper's core numbers don't add up: TP+FN exceeds the total number of phishing emails, so the headline recall drops are internally inconsistent, even though the general direction matches prior work. read the letter →

arxiv 2411.13874 v1 pith:HQYFOXWJ submitted 2024-11-21 cs.CR cs.AI

Khalifa Afane , Wenqi Wei , Ying Mao , Junaid Farooq , Juntao Chen This is my paper · ORCID
classification cs.CRcs.AI
keywords phishingdetectionlargelanguagemodelszero-shotpromptingfew-shotsemanticevasiondataaugmentationemailsecurityadversarialrewriting
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that current phishing defenses—commercial filters, classic machine-learning classifiers, and even LLM-based detectors—lose significant accuracy when phishing emails are rewritten by an LLM to sound more legitimate. Using GPT-4o with zero-shot and few-shot prompts to rephrase emails from two public datasets, the authors measure recall drops across every detector, with the largest drops on few-shot rephrased emails (e.g., Gmail recall from 98.12% to 83.04% on the Nazario dataset). The paper also claims a remedy: training models on an LLM-augmented dataset of 5,000 emails improves detection of rephrased attacks by roughly 10 to 13 accuracy points. If right, this means simple prompt-based rewriting is enough to lower the bar for phishing attacks, and current defenses need updating or augmentation to keep up.

What carries the argument

The machinery is the rephrasing-plus-evaluation pipeline. Zero-shot prompting hands GPT-4o a single instruction to rephrase the email with the same sender, receiver, subject, and body, avoiding urgency, deadlines, generic greetings, and money amounts while still asking for the same information; few-shot prompting adds three crafted examples of rephrased phishing emails. The rephrased emails are then passed through the same detector battery as the originals—Gmail Spam Filter, SpamAssassin, Proofpoint, SVM, logistic regression, naive Bayes, and five LLMs—and the resulting accuracy, precision, recall, and F1 scores are compared. The same LLM-generated outputs are used to construct the LLM-Nazario augmented training set, turning the attack generator into a data-augmentation tool.

What would settle it

Manually inspect the 600 rephrased Nazario emails (or a random sample) and check whether each still contains a working phishing URL or a concrete request for credentials; recompute the detectors' recall on only those emails whose payload survived, and if the gap narrows to near zero, the paper's central claim would be refuted as an artifact of rewriting.

Watch

Extended reading notes

Core claim

Stated on the paper's own terms, the discovery is that a single prompt-based rewriting step—asking GPT-4o to rephrase a phishing email so it sounds more legitimate while keeping the same core topic—is enough to push phishing emails across the detection boundary of every system tested. On the Nazario dataset, Gmail's recall on phishing falls from 98.12% to 83.04% for few-shot rephrased emails, and naive Bayes falls from 94.43% to 75.44%; the same pattern holds on the Nigerian Fraud dataset, where Gemini's recall drops to 75.64%. The paper attributes this to a shift in vocabulary: suspicious markers like 'urgent' and 'large payment' are replaced by neutral terms like 'account update' and 'credentials', narrowing the decision boundary between phishing and legitimate mail. It then claims that retraining SVM, naive Bayes, and logistic regression on an LLM-augmented dataset (LLM-Nazario, 5,000 emails generated by GPT-4 and Llama 3) improves accuracy on rephrased emails by 12.86, 10.38, and 10.90 percentage points respectively.

Load-bearing premise

The rephrased emails still contain the same phishing payload—the malicious link or the request for credentials—so the measured detection gap reflects evasion of real attacks rather than a change in content.

Editorial extensions

If this is right

  • Commercial filters and machine-learning classifiers will miss a larger share of LLM-rephrased phishing than of conventional phishing, so security teams should treat style-based keyword signals as a weak link.
  • Training the same classical classifiers on LLM-augmented data recovers a meaningful share of the lost accuracy, giving defenders a low-cost update path that does not require changing the model family.
  • LLM-based detectors also lose recall on rephrased email, which means the detection gap is not specific to keyword-based tools; any detector that relies on surface text is vulnerable to style rewriting.
  • Because few-shot rewriting consistently produces the largest recall drops, an attacker with only a handful of examples can make phishing emails substantially harder to catch without fine-tuning.
  • The released LLM-Nazario dataset is intended to serve as a shared resource for fine-tuning and evaluating future phishing detectors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct check the paper leaves undone is verifying that the rewritten emails retain the original phishing payload; if many rewrites drop the link or credential request, the reported recall gap overstates the security risk, and a filtered re-analysis on payload-preserving rewrites would settle this.
  • The decision-boundary mechanism suggests an arms race: as detectors train on the current set of LLM-rephrased variants, attackers can iterate prompts to generate new neutral-sounding phrasings, so evaluations should include adaptive, closed-loop rewriting rather than a static test set.
  • The results are based on English text from two public datasets with sender, receiver, subject, and body features; they may not transfer to non-English phishing, image-based emails, or messages where the malicious content lives in attachments or links rather than the body.
  • If LLM-generated bulk mail were required to carry model provenance metadata, detectors could use that signal—but attackers could simply strip it, so the regulatory value depends on enforcement at the email-service level rather than on content analysis alone.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper evaluates how well traditional phishing detectors (Gmail, SpamAssassin, Proofpoint), machine learning classifiers (Naive Bayes, SVM, Logistic Regression), and five LLMs detect phishing emails after the emails are rephrased by GPT-4o and Llama 3 using zero-shot and few-shot prompting. It reports notable recall declines on rephrased emails and proposes using LLM-rephrased emails as data augmentation to improve phishing detection. The authors introduce a new dataset, LLM-Nazario, for this purpose.

Significance. The research question is timely and practically important: if LLM-rephrased phishing emails systematically evade current detectors, this is a serious security concern, and the proposed data-augmentation direction could be valuable for defense. The paper also covers a broad set of detectors and two phishing corpora, which is commendable. However, the quantitative evidence is undermined by fundamental arithmetic inconsistencies in the confusion-matrix tables, a missing verification that rephrased emails preserve the phishing payload, and a circular data-augmentation evaluation. Because the central empirical claim rests entirely on the flawed tables, the results as presented are not reliable. If corrected and validated, the paper could make a meaningful contribution, but in its current form the findings do not support the conclusions.

major comments (4)
  1. [Section IV-A, Table I] The confusion-matrix counts are arithmetically impossible. The paper states that the Nazario test set contains 600 phishing and 600 legitimate emails, so TP+FN must equal 600 for any fixed test set. Yet the few-shot Gmail row reports TP=573 and FN=117, giving TP+FN=690, and the zero-shot row gives 614; the same problem appears in Table III (e.g., Gmail zero-shot on Nigerian Fraud gives TP+FN=436 versus the stated 400 phishing emails). The reported recalls such as 83.04% equal TP_original/(TP_original+FN_rephrased), which mixes counts from different evaluation corpora. All metrics derived from these tables are therefore invalid, and the paper's central claim that detection accuracy declines on rephrased emails is unsupported by the reported data.
  2. [Section III-C] The paper does not verify that the LLM-rephrased emails preserve the phishing payload. The zero-shot prompt instructs the model to "ask for the same information the original email is asking for," but there is no check that the malicious link, credential request, or harmful instruction survives rewriting. If the rewrites drop the payload, then the observed detection gap measures style change rather than a real security risk. The authors should provide evidence—either manual inspection or automated checks—that the rephrased emails still constitute phishing attempts before claiming a security-relevant evasion.
  3. [Section IV-C and Figure 5] The data-augmentation evaluation is circular. The models are trained on GPT-4- and Llama-3-rephrased emails and then tested on LLM-rephrased emails generated by the same model family, leading to improvements such as SVM's 12.86% accuracy gain. With no held-out distribution or cross-model testing, the observed gains may simply reflect train/test overlap in wording and style, not improved generalization to novel LLM-generated attacks. A valid evaluation should test on rephrased emails from a different LLM family or on independently generated phishing variants.
  4. [Section III-B and Section IV-B] The experimental setup lacks essential statistical and methodological detail. The paper reports a majority vote over three iterations but gives no variance, confidence intervals, or significance testing, which is insufficient for a claim based on differences between 98.12% and 83.04% recall. In addition, Section IV-B states that machine learning models were trained on 1500 emails from the original datasets, which is inconsistent with the dataset sizes described in Section III-A (1200 for Nazario, 800 for Nigerian Fraud); the train/test split and any overlap between training and evaluation subsets are not clearly described. These omissions prevent the reader from assessing the reliability of the reported results.
minor comments (4)
  1. [Table II caption] The caption of Table II says "Nigerian Fraud Dataset," but Section IV-A describes Table II as showing LLM performance on the Nazario dataset; this inconsistency should be corrected.
  2. [References] References [34] and [35] are the same paper (Kojima et al.) and should be merged.
  3. [Section IV] The Naive Bayes equation is written with a single word w_i as the conditioning event, but the preceding text indicates the probability is conditioned on a word's presence; the notation should be clarified. The logistic regression formula also introduces w and x without explicitly defining the feature vector for an email.
  4. [Whole paper] There are several typos and grammar errors, including "sophisticated models has already outperformed" (Abstract/Introduction), "the the emails" (Section III-B), and "significantl" (Conclusion). A careful language edit is needed.

Circularity Check

1 steps flagged · score 6.0 of 10

Core evasion measurement is non-circular, but the LLM data-augmentation benefit is partially circular because the same model family generates both the training augmentation and the test rephrased emails.

  1. fitted input called prediction [Section IV.C (Using LLMs for Rephrasing and Data Augmentation), Figure 5 paragraph]
    "Figure 5 illustrates the detection accuracy of the SVM, Naive Bayes, and Logistic Regression models, each trained separately on three datasets: the traditional phishing dataset, an LLM-augmented dataset generated by GPT-4, and an LLM-augmented dataset generated by Llama3. ... The models were then tested exclusively on a set of LLM-rephrased phishing emails, crafted using advanced prompt engineering to evade detection."

    The classifiers are trained on augmented datasets built from GPT-4 and Llama3 rephrased phishing emails and are then evaluated on a test set of LLM-rephrased phishing emails. The paper does not state that the test emails were generated by a different, held-out LLM or an independent attack style. In the absence of such separation, the test set lies in the same generative distribution as the training augmentation, so the reported improvement (e.g., SVM reaching 94.40% accuracy, a 12.86% increase) partly measures in-distribution fit to the GPT-4/Llama3 rephrasing distribution rather than general robustness to novel LLM-rephrased attacks.

full rationale

The paper's primary empirical claim—that GPT-4o zero-shot and few-shot rephrasing degrades detection by Gmail, SpamAssassin, Proofpoint, ML classifiers, and LLM detectors—is not circular in the derivation-chain sense. The rephrasing process is independent of the detectors under test, and the reported drops are presented as measured outcomes rather than as consequences of the evaluation setup. No self-citation chain or imported uniqueness theorem is load-bearing. The one genuine circularity concern is in the data-augmentation contribution: models are trained on LLM-rephrased emails generated by GPT-4 and Llama3 and then evaluated on LLM-rephrased emails, with no evidence of a held-out generator or distinct attack style, so the improved accuracy partly reduces to fitting the same generative distribution. I also note that Tables I and III contain an arithmetic inconsistency—TP counts are identical across original, zero-shot, and few-shot rows while FN grows, so TP+FN sometimes exceeds the total number of phishing emails (e.g., Gmail few-shot: 573+117=690 against 600 phishing emails). That is a serious reproducibility/measurement flaw, but it is a correctness issue, not a circular derivation, so it does not by itself raise the circularity score.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The numerical results depend on several hand-chosen experimental settings (sample sizes, number of examples, encoding method) and on unverified domain assumptions about dataset representativeness and the preservation of phishing intent. No new physical or conceptual entities are introduced.

free parameters (4)
  • Subsampled dataset sizes = 1,200 Nazario emails; 800 Nigerian Fraud emails
    The authors selected these sample sizes from larger public datasets, and all reported metrics depend on this choice. No sampling strategy or representativeness analysis is given.
  • Number of few-shot examples = 3
    The few-shot rephrasing prompt uses three example emails. The choice is arbitrary and affects the style of the generated phishing variants.
  • Training set size for ML models = 1,500 emails (mentioned in Section IV-B)
    The machine learning models were trained on 1,500 emails from the original datasets. The composition and split are not fully specified.
  • TF-IDF encoding = Selected because it outperformed BoW and Word2Vec by 2.6% and 5.4% respectively
    The encoding method was chosen after observing its relative performance on the data, which introduces a selection effect into the reported ML results.
assumptions (4)
  • domain assumption The Nazario and Nigerian Fraud datasets are representative of real-world phishing and legitimate email.
    The validity of all detection measurements depends on these public datasets reflecting realistic phishing content and benign inbox traffic. The paper does not justify this representativeness.
  • domain assumption LLM rephrasing with the provided prompts preserves the original phishing intent and payload.
    The zero-shot prompt instructs the model to ask for the same information and keep the core topic, but no verification, human evaluation, or automatic check confirms that the rewritten emails still contain the malicious ask.
  • domain assumption The binary decisions of Gmail, Proofpoint, and SpamAssassin are treated as ground truth for phishing classification.
    The paper records spam folder placement, quarantine status, and spam flags as definitive labels, which may not account for configuration, account history, or other deployment factors.
  • domain assumption Majority vote over three iterations is sufficient to stabilize the stochastic behavior of LLM detectors.
    The paper uses three runs and a majority vote without reporting variance or testing whether three iterations are enough for stable estimates.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Next-Generation Phishing: How LLM Agents Empower Cyber Attackers." pith.science (2026). https://pith.science/paper/HQYFOXWJ

@misc{pith2026241113874,
  author       = {Pith},
  title        = {Pith review of: Next-Generation Phishing: How LLM Agents Empower Cyber Attackers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HQYFOXWJ}},
  note         = {Machine review of arXiv:2411.13874}
}
read the original abstract

The escalating threat of phishing emails has become increasingly sophisticated with the rise of Large Language Models (LLMs). As attackers exploit LLMs to craft more convincing and evasive phishing emails, it is crucial to assess the resilience of current phishing defenses. In this study we conduct a comprehensive evaluation of traditional phishing detectors, such as Gmail Spam Filter, Apache SpamAssassin, and Proofpoint, as well as machine learning models like SVM, Logistic Regression, and Naive Bayes, in identifying both traditional and LLM-rephrased phishing emails. We also explore the emerging role of LLMs as phishing detection tools, a method already adopted by companies like NTT Security Holdings and JPMorgan Chase. Our results reveal notable declines in detection accuracy for rephrased emails across all detectors, highlighting critical weaknesses in current phishing defenses. As the threat landscape evolves, our findings underscore the need for stronger security controls and regulatory oversight on LLM-generated content to prevent its misuse in creating advanced phishing attacks. This study contributes to the development of more effective Cyber Threat Intelligence (CTI) by leveraging LLMs to generate diverse phishing variants that can be used for data augmentation, harnessing the power of LLMs to enhance phishing detection, and paving the way for more robust and adaptable threat detection systems.

Figures

Figures reproduced from arXiv: 2411.13874 by the authors.

Figure 1
Figure 1. Evaluation methodology workflow, highlighting the differences in detection effectiveness on average between traditional and LLM-rephrased emails. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Depiction of the decision boundary shift between traditional phishing emails and LLM-rephrased phishing emails in terms of classification probability. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Classification Results for 5 Original Emails by Llama 3 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Classification Results for 5 Rephrased Emails by Llama 3 (Few-Shot Prompting) [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Accuracy Comparison of SVM, Naive Bayes, and Logistic Regression in Detecting Rephrased Emails: Traditional vs. LLM-Augmented Datasets. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 28 canonical work pages

  1. [1]

    IEEE Communications Surveys & Tutorials, 15(4), pp

    Almomani, A., Gupta, B.B., Atawneh, S., Meulenberg, A., Almomani, E.: A Survey of Phishing Email Filtering Techniques. IEEE Communications Surveys & Tutorials, 15(4), pp. 2070–2090. IEEE (2013)

    work page 2013

  2. [2]

    PloS one, 14(1), pp

    Jones, H.S., Towse, J.N., Race, N., Harrison, T.: Email fraud: The search for psychological predictors of susceptibility. PloS one, 14(1), pp. e0209684. Public Library of Science (2019)

    work page 2019

  3. [3]

    Frontiers in Psychology, 11, pp

    McAlaney, J., Hills, P.J.: Understanding phishing email processing and perceived trustworthiness through eye tracking. Frontiers in Psychology, 11, pp. 1756. Frontiers Media SA (2020)

    work page 2020

  4. [4]

    PhD Thesis, The University of St Andrews (2023)

    Wang, Y .: Mitigating phishing threats. PhD Thesis, The University of St Andrews (2023)

    work page 2023

  5. [5]

    In: 2024 IEEE Interna- tional Conference on Cyber Security and Resilience (CSR), pp

    Tihanyi, N., Ferrag, M.A., Jain, R., Bisztray, T., Debbah, M.: CyberMet- ric: A Benchmark Dataset based on Retrieval-Augmented Generation for Evaluating LLMs in Cybersecurity Knowledge. In: 2024 IEEE Interna- tional Conference on Cyber Security and Resilience (CSR), pp. 296–302. IEEE (2024)

    work page 2024

  6. [6]

    arXiv preprint arXiv:2402.07927 (2024)

    Sahoo, P., Singh, A.K., Saha, S., Jain, V ., Mondal, S., Chadha, A.: A Systematic Survey of Prompt Engineering in Large Language Models: Techniques and Applications. arXiv preprint arXiv:2402.07927 (2024)

    arXiv 2024

  7. [7]

    Language models are unsupervised multitask learners,

    A. Radford, J. Wu, R. Child, D. Luan, D. Amodei, and I. Sutskever, “Language models are unsupervised multitask learners,” OpenAI Blog , vol. 1, no. 8, p. 9, 2019

    work page 2019

  8. [8]

    In: Proceedings of the 16th International Conference on World Wide Web, pp

    Fette, I., Sadeh, N., Tomasic, A.: Learning to detect phishing emails. In: Proceedings of the 16th International Conference on World Wide Web, pp. 649–656. ACM, Banff (2007)

    work page 2007

Show all 38 references
  1. [9]

    In: 2009 Symposia and Workshops on Ubiquitous, Autonomic and Trusted Computing, pp

    Ma, L., Ofoghi, B., Watters, P., Brown, S.: Detecting phishing emails using hybrid features. In: 2009 Symposia and Workshops on Ubiquitous, Autonomic and Trusted Computing, pp. 493–497. IEEE (2009)

    2009

  2. [10]

    In: Computer Security–ESORICS 2012: 17th European Symposium on Research in Computer Security, pp

    Verma, R., Shashidhar, N., Hossain, N.: Detecting phishing emails the natural language way. In: Computer Security–ESORICS 2012: 17th European Symposium on Research in Computer Security, pp. 824–841. Springer, Pisa (2012)

    2012

  3. [11]

    Carnegie Mellon University (2009)

    Sheng, S., Wardman, B., Warner, G., Cranor, L., Hong, J., Zhang, C.: An Empirical Analysis of Phishing Blacklists. Carnegie Mellon University (2009)

    2009

  4. [12]

    Computers & Security, 110, 102414

    Alhogail, A., Alsabih, A.: Applying Machine Learning and Natural Language Processing to Detect Phishing Email. Computers & Security, 110, 102414. Elsevier (2021)

    2021

  5. [13]

    arXiv preprint arXiv:2402.18649 (2024)

    Wu, F., Zhang, N., Jha, S., McDaniel, P., Xiao, C.: A New Era in LLM Security: Exploring Security Concerns in Real-World LLM-based Systems. arXiv preprint arXiv:2402.18649 (2024)

    2024 arXiv

  6. [14]

    High-Confidence Computing, pp

    Yao, Y ., Duan, J., Xu, K., Cai, Y ., Sun, Z., Zhang, Y .: A survey on large language model (LLM) security and privacy: The good, the bad, and the ugly. High-Confidence Computing, pp. 100211. Elsevier (2024)

    2024

  7. [15]

    arXiv 2305 (2023)

    Hazell, J.: Spear phishing with large language models. arXiv 2305 (2023)

    2023

  8. [16]

    arXiv preprint arXiv:2302.13971 (2023)

    Touvron, H., Lavril, T., Izacard, G., Martinet, X., Lachaux, M.-A., Lacroix, T., Rozi `ere, B., Goyal, N., Hambro, E., Azhar, F., et al.: Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971 (2023)

    2023 arXiv

  9. [17]

    M., Hauth, A., Millican, K., et al.: Gemini: A family of highly capable multimodal models

    Team, Gemini, Anil, R., Borgeaud, S., Alayrac, J.-B., Yu, J., Sori- cut, R., Schalkwyk, J., Dai, A. M., Hauth, A., Millican, K., et al.: Gemini: A family of highly capable multimodal models. arXiv preprint arXiv:2312.11805 (2023)

    2023 arXiv

  10. [18]

    [Online]

    Anthropic: Claude. [Online]. Available: https://www.anthropic.com/news/introducing-claude

  11. [19]

    L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al.: GPT-4 Technical Report

    Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F. L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al.: GPT-4 Technical Report. arXiv preprint arXiv:2303.08774 (2023)

    2023 arXiv

  12. [20]

    IEEE Access (2024)

    Heiding, F., Schneier, B., Vishwanath, A., Bernstein, J., Park, P.S.: Devising and detecting phishing emails using large language models. IEEE Access (2024)

    2024

  13. [21]

    Universal spam detection using transfer learning of BERT model,

    Koide, T., Fukushi, N., Nakano, H., Chiba, D.: ChatSpamDetector: Leveraging Large Language Models for Effective Phishing Email De- tection. arXiv preprint arXiv:2402.18093 (2024) V . S. Tida and S. Hsu, “Universal spam detection using transfer learning of BERT model,” arXiv pr...

    2024 arXiv

  14. [22]

    SpamAssassin,

    “SpamAssassin,” Apache, [Online]. Available: https://spamassassin.apache.org/. [Accessed: 10-Sep-2024]

    2024

  15. [23]

    A comparative analysis between technical and non-technical phishing defenses,

    J. C. Brickley, K. Thakur, and A. S. Kamruzzaman, “A comparative analysis between technical and non-technical phishing defenses,” Inter- national Journal of Cyber-Security and Digital F orensics , vol. 10, no. 1, pp. 28–41, 2021

    2021

  16. [24]

    Spam-t5: Benchmarking large lan- guage models for few-shot email spam detection,

    M. Labonne and S. Moran, “Spam-t5: Benchmarking large lan- guage models for few-shot email spam detection,” arXiv preprint arXiv:2304.01238, 2023

    2023 arXiv

  17. [25]

    In: 2024 IEEE 14th Annual Computing and Communication Workshop and Conference (CCWC), pp

    Chataut, R., Gyawali, P.K., Usman, Y .: Can AI Keep You Safe? A Study of Large Language Models for Phishing Detection. In: 2024 IEEE 14th Annual Computing and Communication Workshop and Conference (CCWC), pp. 0548–0554. IEEE (2024)

    2024

  18. [26]

    arXiv preprint arXiv:2302.05733 (2023)

    Kang, D., Li, X., Stoica, I., Guestrin, C., Zaharia, M., Hashimoto, T.: Exploiting Programmatic Behavior of LLMs: Dual-Use Through Standard Security Attacks. arXiv preprint arXiv:2302.05733 (2023)

    2023 arXiv

  19. [27]

    2024 IEEE Symposium on Security and Privacy (SP), pp

    Roy, S.S., Thota, P., Naragam, K.V ., Nilizadeh, S.: From Chatbots to Phishbots?: Phishing Scam Generation in Commercial Large Language Models. 2024 IEEE Symposium on Security and Privacy (SP), pp. 221– 221, IEEE Computer Society (2024)

    2024

  20. [28]

    In: 3rd IEEE International Conference on Computing and Machine Intelligence (ICMI), pp

    Champa, A.I., Rabbi, M.F., Zibran, M.F.: Curated Datasets and Feature Analysis for Phishing Email Detection with Machine Learning. In: 3rd IEEE International Conference on Computing and Machine Intelligence (ICMI), pp. 1–7 (to appear) (2024)

    2024

  21. [29]

    In: 2019 International Engineering Conference (IEC), pp

    Qader, W.A., Ameen, M.M., Ahmed, B.I.: An overview of bag of words; importance, implementation, applications, and challenges. In: 2019 International Engineering Conference (IEC), pp. 200–204. IEEE (2019)

    2019

  22. [30]

    In: Proceedings of the First Instructional Conference on Machine Learning, vol

    Ramos, J.: Using tf-idf to determine word relevance in document queries. In: Proceedings of the First Instructional Conference on Machine Learning, vol. 242(1), pp. 29–48. Citeseer (2003)

    2003

  23. [31]

    Natural Language Engineering, 23(1), pp

    Church, K.W.: Word2Vec. Natural Language Engineering, 23(1), pp. 155–162. Cambridge University Press (2017)

    2017

  24. [32]

    Computer-Aided Civil and Infrastructure Engineering 38(11), 1536–1554 (2023)

    Yong, G., Jeon, K., Gil, D., Lee, G.: Prompt engineering for zero-shot and few-shot defect detection and classification using a visual-language pretrained model. Computer-Aided Civil and Infrastructure Engineering 38(11), 1536–1554 (2023)

    2023

  25. [33]

    arXiv preprint arXiv:2110.08207 (2021)

    Sanh, V ., Webson, A., Raffel, C., Bach, S.H., Sutawika, L., Alyafeai, Z., Chaffin, A., Stiegler, A., Scao, T.L., Raja, A., et al.: Multitask Prompted Training Enables Zero-Shot Task Generalization. arXiv preprint arXiv:2110.08207 (2021)

    2021 arXiv

  26. [34]

    Advances in Neural Information Processing Systems, 35, pp

    Kojima, T., Gu, S.S., Reid, M., Matsuo, Y ., Iwasawa, Y .: Large Lan- guage Models Are Zero-Shot Reasoners. Advances in Neural Information Processing Systems, 35, pp. 22199–22213 (2022)

    2022

  27. [35]

    Advances in Neural Information Processing Systems 35, 22199–22213 (2022)

    Kojima, T., Gu, S.S., Reid, M., Matsuo, Y ., Iwasawa, Y .: Large lan- guage models are zero-shot reasoners. Advances in Neural Information Processing Systems 35, 22199–22213 (2022)

    2022

  28. [36]

    Advances in Neural Information Processing Systems 35, 30378–30392 (2022)

    Ye, X., Durrett, G.: The unreliability of explanations in few-shot prompt- ing for textual reasoning. Advances in Neural Information Processing Systems 35, 30378–30392 (2022)

    2022

  29. [37]

    In: Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems, pp

    Reynolds, L., McDonell, K.: Prompt programming for large language models: Beyond the few-shot paradigm. In: Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems, pp. 1– 7 (2021)

    2021

  30. [38]

    arXiv preprint arXiv:2309.12307 (2023)

    Chen, Y ., Qian, S., Tang, H., Lai, X., Liu, Z., Han, S., Jia, J.: LongLoRA: Efficient Fine-Tuning of Long-Context Large Language Models. arXiv preprint arXiv:2309.12307 (2023)

    2023 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.